Added some reference images off my bench. These images are for educational and reference use ONLY. Taking a look at the mechanical parts, you can see there's a two-position switch that the lever will push as well, for hot and cold dispensing. Is there an electrical difference to the two? The cold dispense doesn't need to reheat the heat-block afterwards, so maybe that's what it's it.

The pump is on the right hand side in the image, along the pipe rising from bottom of the water reservoir. There's a T-junction on the pipe, the other branch going directly the the water-dispenser, the other taking a long detour to the heat-block on the left hand side and back. The mechanical action of the lever blocks either one of these tubes as time along with toggling the two position switch shown. For fully automatic dispensing, the cold tube would need to be blocked, or a mechanical actuator added to choose between the tubes. In that case, a divert could be added as well to allow pre-heating (or cooling) the tubes and the water in them by letting it briefly run off, or back into the water reservoir (which would heat water there, making cold drinks even warmer...). Especially since there's no divert, the tubes could use with some insulation to keep them at desired temperature. Failing both of these, you might want to just let water run off the machine with no capsule inserted for a while before dispensing.

There's a close up of the back-side of the low-voltage circuitry, where you almost make out the writing on the SOIC16 mentioned in the details. R20 hooks up heater, R19 hooks up water pump, the other side of the opto-isolators can be seen going direct to the microcontroller, which might be helpful for a reverse-engineering and re-programming project. But this isn't one. Worst come to worst, cutting up those thin traces going to the microcontroller and hooking up our own controller would let us in direct control. Near L2 at upper left is the hookup for the NTC thermistor though; we can see a lot of discrete components involved, and the traces go under the microcontroller, so it's not easy to figure out how it works right away. That is something I may have to focus on, though.

Finally, there's a picture of the optocouplers from the front so you can see the model and roughly how they line up. It also has J12 marked as ground, so we have our ground and can see it circling the edges of the low volatage side and the hole at the middle. Just a reminder, when that thing is hooked up to the wall, a lot of the connectors will have deadly levels of current on them, so hands off when it's hooked up, and make sure nothing you add or change bridges the upper and lower side of the board, divided through those two optocoupplers and the power-suppply feedback optocoupler in the middle.